1 / 30

Quaternion and Virtual Trackball

Quaternion and Virtual Trackball. CSE 781 Introduction to 3D Image Generation Han-Wei Shen Winter 2007. y. y. y. x. x. x. z. z. z. Euler Rotation Problems. Gimbal Lock – lose one degree of freedom Problem happens when the axes of rotation line up on top of each other. For example: .

jaden
Download Presentation

Quaternion and Virtual Trackball

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Quaternion and Virtual Trackball CSE 781 Introduction to 3D Image Generation Han-Wei Shen Winter 2007

  2. y y y x x x z z z Euler Rotation Problems • Gimbal Lock – lose one degree of freedom • Problem happens when the axes of rotation line up on top of each other. For example: Step one: Rotate(0, 0,0,1) Step2: Rotate(90, 0,1,0) Step 3: Rotate(??, 0,0,1) This is same as rotation x !! 

  3. y x Euler Rotation Problems • Rotations with Euler angles to change from one orientation to another are not unique. Example: (x,y,z) rotation to achieve the following: R R x y y OR z z x z x R z y R z y Rotate(180, 1,0,0) Rotate(180, 0,1,0) then Rotate(180,0,0,1) Euler angles: (0,0,0) -> (180,0,0) Euler angles: (0,0,0) -> (0,180,180)

  4. Quaternion • Invented in 1843 as an extension to the complex numbers • Used by computer graphics since 1985 • Quaternion: • Provide an alternative method to specify rotation • Can avoid the gimbal lock problem • Allow unique, smooth and continuous rotation interpolations

  5. Mathematical Background • A quaternion is a 4-tuple of real number, which can be seen as a vector and a scalar Q = [qx, qy, qz, qw] = qv + qw, where qw is the real part and qv = iqx + jqy + kqz = (qx, qy, qz)is the imaginary part • i*i = j*j = k*k = -1; • j*k= -k*j= i; k*i=-i*k=j; i*j=-j*i= k; • All the regular vector operations (dot product, cross product, scalar product, addition, etc) applied to the imaginary part qv

  6. Basic Operations • Multiplication: QR = (qv x rv+ rwqv+ qwrv, qwrw - qv.rv) • Addition: Q+R = (qv+rv, qw+rw) • Conjugate: Q* = (-qv, qw) • Norm (magnitude) = QQ* = Q*Q = qx*qx+qy*qy+qz*qz+qw*qw • Identity i = (0,1) • Inverse Q = (1/ Norm(Q)) Q* • Some more rules can be found in the reference book (real time rendering) pp46 Imaginary real -1

  7. Polar Representation • Remember a 2D unit complex number cosq + i sinq = e • A unit quaternion Q may be written as: Q = (sinfuq , cosf) = cosf + sinfuq, where uq is a unit 3-tuple vector • We can also write this unit quaternion as: Q = e iq uqf

  8. Quaternion Rotation • A rotation can be represented by a unit quaternion Q = (sinfuq, cosf) • Given a point p = (x,y,z) -> we first convert it to a quaternion p’ = ix+jy+kz+ 0 = (pv, 0) • Then, Qp’Q is in fact a rotation of p around uq by an angle 2f !! -1

  9. Rotation Concatenation • Concatenation is easy – just multiply all the quaternions Q1, Q2, Q3, …. Together • There is a one-to-one mapping between a quaternion rotation and 4x4 rotation matrix. (Q3 (Q2 ( Q1 P’ Q1 ) Q2 ) Q3 ) = (Q3*Q2*Q1) P’ (Q1*Q2*Q3 ) -1 -1 -1 -1 -1 -1

  10. Quaternion to Rotation Matrix • Given a quaternion w + xi +yj + kz, it can be translated to the rotation matrix R: 1-2y^2-2z^2 2xy+2wz 2xz-2wy R = 2xy-2wz 1-2x^2-2z^2 2yz+2wx 2xz+2wy 2yz-2wx 1-2x^2-2y^2 • Also you can convert a matrix to quaternion (see the reference book for detail)

  11. Interpolation of Rotation • Should avoid sudden change of orientation and also should maintain a constant angular speed • Each rotation can be represented as a point on the surface of a 4D unit sphere • Need to perform smooth interpolation along this 4D sphere R How to interpolate A and B to get R? A B

  12. Interpolation Rotation • Spherical Linear Interpolation (slerp): Given two unit quaternion (i.e., two rotations), we can create a smooth interpolation using slerp: slerp(Q1, Q2, t) = sin (f(1-t)) sin(ft) sinf sinf where 0<=t<=1 • To compute f, we can use this property: cosf = Q1xQ2x+Q1yQ2y+Q1zQ2z+Q1wQ2w Q1+ Q2

  13. glRotatef(q, 0,1,0) glRotatef(q, 1,0,0) OpenGL - glRotatef(q, 0,0,1) 3D Rotations with Euler Angles • A simple but non-intuitive method – specify separate x, y, z axis rotation angles based on the mouse’s horizontal, vertical, and diagonal movements cos(q) -sin(q) 0 0 sin(q) cos(q) 0 0 0 0 1 0 0 0 0 1 cos(q) 0 sin(q) 0 0 1 0 0 -sin(q) 0 cos(q) 0 0 0 0 1 1 0 0 0 0 cos(q) -sin(q) 0 0 sin(q) cos(q) 0 0 0 0 1

  14. y y x y x x z z Euler Rotation Problems • Gimbal Lock – lose one degree of freedom • Problem happens when the axes of rotation line up on top of each other. For example: z Step one: Rotate(0, 1,0,0) Step2: Rotate(90, 0,1,0) Step 3: Rotate(??, 0,0,1) This is same as rotation x !! 

  15. 3D Rotations with Trackball • Imagine the objects are rotated along with a imaginary hemi-sphere

  16. Virtual Trackball • Allow the user to define 3D rotation using mouse click in 2D windows • Work similarly like the hardware trackball devices

  17. Virtual Trackball • Superimpose a hemi-sphere onto the viewport • This hemi-sphere is projected to a circle inscribed to the viewport • The mouse position is projected orthographically to this hemi-sphere z y (x,y,0) x

  18. z y x Virtual Trackball • Keep track the previous mouse position and the current position • Calculate their projection positions p1 and p2 to the virtual hemi-sphere • We then rotate the sphere from p1 to p2 by finding the proper rotation axis and angle • This rotation ( in eye space!) is then applied to the object (call the rotation before you define the camera with gluLookAt()) • You should also remember to accumulate the current rotation to the previous modelview matrix

  19. z y x Virtual Trackball • The axis of rotation is given by the normal to the plane determined by the origin, p1 , and p2 • The angle between p1 and p2 is given by n = p1  p1 | sin q| =

  20. z y (x,y,0) x Virtual Trackball • How to calculate p1 and p2? • Assuming the mouse position is (x,y), then the sphere point P also has x and y coordinates equal to x and y • Assume the radius of the hemi-sphere is 1. So the z coordinate of P is • Note: normalize viewport y extend to -1 to 1 • If a point is outside the circle, project it to the nearest point on the circle (set z to 0 and renormalize (x,y))

  21. Virtual Trackball Visualization of the algorithm

  22. Example • Example from Ed Angel’s OpenGL Primer • In this example, the virtual trackball is used to rotate a color cube • The code for the colorcube function is omitted • I will not cover the following code, but I am sure you will find it useful

  23. Initialization #define bool int /* if system does not support bool type */ #define false 0 #define true 1 #define M_PI 3.14159 /* if not in math.h */ int winWidth, winHeight; float angle = 0.0, axis[3], trans[3]; bool trackingMouse = false; bool redrawContinue = false; bool trackballMove = false; float lastPos[3] = {0.0, 0.0, 0.0}; int curx, cury; int startX, startY;

  24. The Projection Step voidtrackball_ptov(int x, int y, int width, int height, float v[3]) { float d, a; /* project x,y onto a hemisphere centered within width, height , note z is up here*/ v[0] = (2.0*x - width) / width; v[1] = (height - 2.0F*y) / height; d = sqrt(v[0]*v[0] + v[1]*v[1]); v[2] = cos((M_PI/2.0) * ((d < 1.0) ? d : 1.0)); a = 1.0 / sqrt(v[0]*v[0] + v[1]*v[1] + v[2]*v[2]); v[0] *= a; v[1] *= a; v[2] *= a; }

  25. glutMotionFunc (1) Void mouseMotion(int x, int y) { float curPos[3], dx, dy, dz; /* compute position on hemisphere */ trackball_ptov(x, y, winWidth, winHeight, curPos); if(trackingMouse) { /* compute the change in position on the hemisphere */ dx = curPos[0] - lastPos[0]; dy = curPos[1] - lastPos[1]; dz = curPos[2] - lastPos[2];

  26. glutMotionFunc (2) if (dx || dy || dz) { /* compute theta and cross product */ angle = 90.0 * sqrt(dx*dx + dy*dy + dz*dz); axis[0] = lastPos[1]*curPos[2] – lastPos[2]*curPos[1]; axis[1] = lastPos[2]*curPos[0] – lastPos[0]*curPos[2]; axis[2] = lastPos[0]*curPos[1] – lastPos[1]*curPos[0]; /* update position */ lastPos[0] = curPos[0]; lastPos[1] = curPos[1]; lastPos[2] = curPos[2]; } } glutPostRedisplay(); }

  27. Idle and Display Callbacks void spinCube() { if (redrawContinue) glutPostRedisplay(); } void display() { glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT); if (trackballMove) { glRotatef(angle, axis[0], axis[1], axis[2]); } colorcube(); glutSwapBuffers(); }

  28. Mouse Callback void mouseButton(int button, int state, int x, int y) { if(button==GLUT_RIGHT_BUTTON) exit(0); /* holding down left button allows user to rotate cube */ if(button==GLUT_LEFT_BUTTON) switch(state) { case GLUT_DOWN: y=winHeight-y; startMotion( x,y); break; case GLUT_UP: stopMotion( x,y); break; } }

  29. Start Function void startMotion(int x, int y) { trackingMouse = true; redrawContinue = false; startX = x; startY = y; curx = x; cury = y; trackball_ptov(x, y, winWidth, winHeight, lastPos); trackballMove=true; }

  30. Stop Function void stopMotion(int x, int y) { trackingMouse = false; /* check if position has changed */ if (startX != x || startY != y) redrawContinue = true; else { angle = 0.0; redrawContinue = false; trackballMove = false; } }

More Related